Comparison of Random Search Method and Genetic Algorithm for Stratification

Abstract

This note shows that the genetic algorithm proposed by Keskintürk and Er (Computational Statistics and Data Analysis 2007, 52, 53–67) usually provides similarly effective stratification to that of the random search algorithm proposed by Kozak (Statistics in Transition 2004, 6(5), 797–806), although in some situations it can be noticeably less effective. Despite this, it is suggested that quite likely genetic algorithms have potential to be a means of very efficient stratification, especially in complex stratification problems.

Keywords:

• Survey sampling
• Global optimization
• Stratified sampling

Mathematical Subject Classification:

• Primary 62D05
• Secondary 62P20
• 62P25

Acknowledgments

I would like to thank Drs Keskintürk and Er for making the populations data used in this paper available via Internet and for their personal agreement to use them in my research.